NOOTROPIC
000%synthesizing skills…
tap to skip
Methodology

How the numbers are made.

No black boxes: how house skills reach a verdict, and what the leaderboard and payment feed actually count.

01

Deterministic, no paid AI

House skills are hand-written rules and parsers built from public security research and open-source tooling (credited on each skill page). The same input always gives the same answer, in milliseconds, at near-zero cost.

02

Risk scores

Each rule that fires carries a weight between 0 and 1. Weights combine as independent signals: score = 1 − Π(1 − wᵢ), shown as 0–100. Text found in a hidden layer (invisible Unicode, base64, CSS-hidden HTML) gets a higher weight, because nobody hides benign text.

03

Verdicts

Injection Shield: clean below 25, suspicious from 25, malicious from 60. Phish Lens: low-risk below 35, suspicious from 35, dangerous from 70; official brand domains are capped at trusted. Calldata X-Ray reports the highest severity of any flag.

04

Leaderboard & feed

Counts come from calls that pass through Nootropic: house-skill calls and paid calls relayed to listed sellers. Free sample doses count as calls but not as paid calls or volume. Calls an agent makes directly to a seller's endpoint are not visible to us and are not counted.

05

Simulated vs. on-chain

Until mainnet payments go live, house-skill settlement is simulated: signatures are verified for real, but no funds move, and those payments are labelled “simulated” everywhere. Payments to listed sellers settle on-chain through the seller's own infrastructure.